Install

Install from the Chrome Web Store and try it out!

Permissions

1. Go to chrome://extensions/?id=jldogdgepmcedfdhgnmclgemehfhpomg
2. Enable "Pin to toolbar"
3. Enable "Allow access to file URLs" to use on local PDF"s.

Web Agent

The Web Agent allows you to directly interact with any webpage to extract data, complete actions, or answer your questions. There are 3 core functions within Web Agent: Seek, Explore, and Extract.

Seek

Chat with our AI agent directly to get specific data points from a webpage, or command it to complete an action.
For example, while on LinkedIn you can command the agent to "Find rtrvr.ai's LinkedIn page and follow the company".

Explore

Crawl entire websites and extract structured data from every page.
For example, if you are on an Amazon product listing page you can command the agent to extract the name, price, sale discount for each product.

• Paginated Lists: "For each of the "YC Partners" go to their LinkedIn profile page and retrieve their name, headline, job, college"
• Folder Directory: "For every PDF file in current directory, explore them and extract paper title and authors"

Export

Structure data from any webpage and export it to Google Sheets with a single click.
For example, while on the TechCrunch Disrupt Speaker page you can command the agent to extract the name, role, and company of every speaker.

Across Tabs

Perform actions across a selection of open tabs with actions such as: Ask Tabs, Compare Tabs, Extract Tabs.

Ask Tabs

Select from open tabs and ask a question or give a command to perform tasks with cross-tab context. For example, you can fill out a job application tab using a LinkedIn tab as context.

Compare

Automatically compare a selection of open tabs.

Extract Tabs

Perform actions and extract arbitrary information from selected open tabs using natural language.

Sheets Workflow

Automate data extraction from multiple websites using a Google Sheet column of URLs. rtrvr.ai can open these URLs as new tabs and perform bulk data extraction. A sample demo output can be found here. You can also construct multi-step workflows, effectively representing a DAG workflow, where each step leverages the output of the previous one.

Example: Given a column of LinkedIn profile URLs, rtrvr.ai can extract information and even write intro emails back to the sheet.

Function Calling

Embed function calls in your prompt using the @ notation (e.g., @querySnowflake()) to add the function's output to the prompt. Alternatively, you can have the LLM construct the function call and parameters to execute (e.g., "Send a Slack message with the summary of this page").

GraphBot

Call @graphBot with a prompt describing the graph you want to render, along with data from the current tab or a selected sheet.

Recordings

Add recordings to guide and ground the model in performing a specific task.

Sheet Context

Use a Google Sheet as context to perform tasks.

Scheduler

Automate tasks to run in the background by entering a cron schedule and selecting a previously successful workflow.

Settings

Access and modify your account settings, payment settings, preferred language model, and more.

Clear Messages in Seek

In Web Agent, Seek: Clear messages for a cleaner context sent to the LLMs as the entire chat history is sent for Seek actions.

Cancel Ongoing Request/Workflow

Close the rtrvr.ai's Side Panel to cancel/stop any ongoing workflows or requests.